Virtual replication of RomeroUK motor/generator by 3D FEM modeling

[teslaalset]

This thread is meant for discussions on simulations regarding the RomeroUK motor / generator.
Please use this thread to share your understanding and discuss disagreements with patience and politeness.
My purpose is to build up knowledge by sharing and converge insights.

As for myself I will share my 'virtual replication' using Ansys Maxwell.
My simulations are done in 3D.
An overview of my own model pictures are attached below.
I copied as much details as possible as shared by RomeroUK, so yes even the washers are in, as well as the ferrite cores.

I will do modelling request on rare occasions, since this will take a lot of time, Maxwell needs several hours to calculate a singe simulation of one complete rotation revolution.

Please, do not ask for the source files, as I will not make them available for open sourcing.
If this one appears to be successful and you need commercial design consultancy, drop me a PM or e-mail.

Next, I will start sharing simulation results of the basic generator principles that I simulated first to build up knowledge.

Enjoy this discussion thread!

Teslaalset.

[teslaalset]

Attached the outline of the model I used for building up basic understanding of the generator coil principle.
There's a fixed magnet positioned above a coil with a ferrite core and a moving magnet beneath it.
The movement is along and within the transparent square bar of the picture.
I generated a nice flux vector simulation represented in an animated GIF file, the file can be downloaded from here:
http://www.multiupload.com/9ZG6AGYPL1

The circuit that is connected to the coil is depicted below.
It's basically a diode bridge and a big capacitor with a parallel load resistor of 100 ohms

Next step, I will show some force graphs of three simulations I ran with different distances of the stationary bias magnet.

[teslaalset]

Some graphs on magnet forces in the direction of the movement of the single coil model.
I varied the distance of the bias magnet by 0.05 and 0.1 cm after starting at position 0.
Position zero is already a rough optimum.

The first three graphs below show the force as function of time.
The total run is 9 ms, where 4.5 ms is where the moving magnet is at TDC, so just centrered below the coil/core.
The graphs are somewhat noisy due to the mesh size of the model.
Mesh size determines the accuracy of the model.
Very fine mesh size will result in perfect graphs, but it will take endless amount of calculation time.
So, I took a mesh size that shows the trend but also gave me acceptable simulation times.

Target is to have the average force over time around zero Newton so effectively no nett negative force (drag) is generated by BEMF when a magnet passes by a coil/core.
You'll notice that two experiments show such results, the third result is mainly resulting into negative values, causing unwanted drag.
To give a more accurate determination, I have integrated those three graphs over time.
This is the fourth graph. You need to look at the end value of the integrated curves. If the end result is zero, the bias allignment is perfect (red line).
The green line clearly shows negative integral force, this will cause BEMF drag and slow down the rotor.

Last graph is the voltage over the load capacitor.
The simulation starts with a bias voltage of 20 volt on the capacitor as a starting point.
You will notice three time phases:
- the start period, where mainly discharge of the capacitor is happening due to the 100 ohms in series
- the load period, where inducted flux charges the capacitor, superseding the discharge by the 100 ohms.
- the last period, where discharge is happening again due to lack of induced flux.
The three graphs represent the various bias offset again of the stator magnets.
So, not only influences the bias distance of the stator magnet the BEMF forces, it also influences the charge capability of the setup.
In this example, you'll notice that the capacitor will discharge in all three occasions, so no power left to drive the motor coils.
By decreasing the distance between the moving magnet and the ferrite core, induced flux can be increased so the result will give a higher end voltage at the end of the total simulation period.
In this particular run, when the offset of the capacitor at the start of the run is chosen at a lower value, say 10 Volt, the same run will show surplus of the capacitor voltage at the end of the same simulation run.

[e2matrix]

Thanks for you help and explanation of this.  Good to know BEMF is one of the key components in the OU generation. 

  Is the Ansys Maxwell program a standalone program or just a component of a larger package needed to do these sims? 

[teslaalset]

Thanks for you help and explanation of this.  Good to know BEMF is one of the key components in the OU generation. 

  Is the Ansys Maxwell program a standalone program or just a component of a larger package needed to do these sims?

Ansys Maxwell is a stand alone program that can be linked to other simulation software of Ansys (so, its compatible with those).
Keep in mind Maxwell is extremely expensive, around 50k$.
It's the best 3D software available for these kind of simulation in my view.
Most of the high tech mechatronics companies use it for professional modeling.